The present invention relates to a noise reduction apparatus. More specifically, the present invention relates to a noise reduction apparatus for reducing sudden sound while leaving a voice signal in an environment where the sudden sound is periodically included in the voice signal.
In related art, there has been a problem that when voice communication is performed using a mobile communication device or the like such as a cellular phone and radio equipment, ambient noise generated in a surrounding environment is mixed in voice, which is original target sound, thereby making it difficult for a receiver to hear the voice. Therefore, a noise reduction process has been widely used as a method of ensuring clarity of the voice. A typical method of the noise reduction process is an adaptive noise reduction process using an adaptive filter. The adaptive filter sequentially changes a characteristic of a filter for reducing the noise in accordance with a change in an environment. That is, the adaptive filter is characterized by being adaptable to an environment. More specifically, by sequentially changing a filter coefficient according to place and time while operating the filter to eliminate only noise components that change according to the place and the time, the adaptive filter is adapted to an environment and ambient noise can be reduced.
There are various kinds of ambient noise. For example, an oxygen cylinder carried by a firefighter at a site of a fire has a function to alert a wearer by vibration sound when a tank capacity becomes less than or equal to a predetermined value. Further, operating noise of a ground compactor is generated at a construction site. Such vibration sound and operating sound is persistent and periodic sudden sound. When radio equipment or the like is used in a state where the sudden sound is generated, the sudden sound is mixed in the radio equipment or the like together with the voice, making it extremely difficult for a receiver to hear the voice.
In order to use the above-mentioned adaptive noise reduction process for reducing the persistent and periodic sudden sound, it is necessary to constantly operate an adaptive signal processing circuit (an adaptive filter circuit). This increases the number of taps in the circuit, thereby increasing the circuit size.
In regard to this matter, Japanese Unexamined Patent Application Publication No. 2002-251823 discloses a method of reducing noise generated in a lens moving part of a camcorder, noise when a head comes into contact with or moves away from a magnetic tape, sound of an optical shutter of a camcorder having a silver salt film camera function, and sudden sound generated during, for example, head seeking at the time of recording on a rotating recording medium. To be more specific, the method blocks signals only in a section where the sudden sound is generated and interpolates lacking information in a blocked section based on voice signals in at least one of before and after the blocked section.
A method disclosed in Japanese Unexamined Patent Application Publication No. 2011-205598 calculates an envelope of a signal including sudden noise, extracts a signal component of the sudden noise, and reduces only the signal component of the sudden noise from the signal including the sudden noise.
In the method disclosed in Japanese Unexamined Patent Application Publication No. 2002-251823, if the signals included in the blocked section is mostly only the sudden sound, a desirable effect can be achieved. However, when the sudden sound is superimposed in the voice section, the voice cannot be appropriately interpolated, bringing a feeling of strangeness to the receiver side.
Further, the method disclosed in Japanese Unexamined Patent Application Publication No. 2011-205598 reduces the signal components specific to the sudden sound (frequency components greater than or equal to 4 kHz not including voice components). The above-mentioned persistent and periodic sudden sound such as the vibration sound or the like usually have components dispersed over various frequencies which overlap with frequency components of voice. It is therefore difficult to detect the sudden sound by this method, and when the sudden sound is reduced, the voice may also be reduced by this method.
In voice communication using a mobile communication device or the like, it is important that a receiver can easily hear voice. However, the present inventor has found a problem that in the above-mentioned conventional sudden sound reduction and interpolation methods, voice with a feeling of strangeness is generated.